BCNet
Exploring Methods For Integrating Data On Socio-Economic And Environmental Processes That Influence Land Use Change: A Pilot Project

(click here for pictures)

I.Introduction: The Pilot Project
II.Methodology

A. Conceptual Framework

B. Methodology for Assessing and Explaining Land Use and Land Cover Change

C. Socio-Economic Analysis
IIIResults

A. Land Cover Changes

B. Explanation of Land Cover Changes

C. Summary
IV.Methodological Implications of the Study

A. A Coceptual Framework

B. A Historical Perspective

C. Information Needs

D. Analytical Mehtods

E. Conclusion

Bibliography



Jennifer M. Olson1, Gichana Manyara, David J. Campbell, David P. Lusch and Jue Hu. Rwanda Society Environment Project, Department of Geography, Michigan State University, East Lansing, MI 48824

February 1995

I. INTRODUCTION: The Pilot Project
The study seeks to inform the Biodiversity Support Program on appropriate conceptual and methodological approaches to understanding society-environment interaction as it relates to biodiversity issues. Effective policy for the conservation of biodiversity cannot view conservation as an end in itself, separate from the societal context in which it is proposed. In order to assess the role of biodiversity conservation in development it is necessary to understand the complex ways in which environmental and societal processes interact.

This Pilot Project was designed to provide insights to appropriate ways of examining such interactions through a preliminary analysis of society-forest interactions in communities adjacent to the Nyungwe Forest Reserve in Rwanda. (click here for pictures)

The Nyungwe Forest Reserve is a high-altitude, tropical montane forest in Southwest Rwanda straddling the Zaire-Nile watershed divide.(click here for pictures) Together with neighboring Kibura National Park in Burundi, it forms one of the largest tropical montane forests in Africa. The forest contains a wide variety of ecosystems, from marshes to bamboo groves to dense forest. Although perhaps best known for its colobus monkeys, it also contains 12 other types of primates, at least 275 bird species, many endemic, over 100 species of orchids, and a wide variety of other plant and animal species. The biodiversity is particularly rich and important because the high-altitude ridge served as a refuge for and source of recolonization of forest plants and animals during and after the drying of much of East Africa during the Ice Ages. Also, the forest is situated in an overlap region between several large-scale biogeographical zones and therefore contains species originating from Tanzania, Ethiopia and the Zaire Basin (Offutt et al 1990).

This unique wealth of biodiversity is situated in the most densely populated and one of the poorest countries in Africa, Rwanda. The study examined the changes in land use and land cover in the areas of the forest and adjacent to it and looked for explanations for these changes in the processes of national and local development.

The study uses a variety of to address the following questions:
-has there been a change of forest cover (augmentation/depletion)? If so, where and what type of cover (forest, woodlot, trees on farms)? what has replaced forest/ what has forest replaced?
-what processes lie behind the changes?
-what are the implications for the protected forest? For local communities?

II. METHODOLOGY
The objective of the Project is to explore methods for integrating data on socio-economic and environmental processes that influence land use change. The principal methods for assessing land use change in the pilot project area are interpretation of aerial photography and analysis of MSS LANDSAT imagery. Societal processes that have contributed to such change include those that have occurred within agricultural communities in the area, government policies and donor activity, and the interaction between these.

While different methods provide insight into each of these processes, integration of the information requires a conceptual framework that leads the researcher to examine the complex web of linkages that determine the interactions between society and natural resource management. These interactions occur between different scales (global, national, local) over time and space and so researchers have to explore the evolution of these interactions over time both at the local level, and between rural communities and national, and even global, forces.

II. A. Conceptual Framework
The conceptual framework adopted by this project is Regional Political Ecology (RPE) (Blaikie and Brookfield 1987; Campbell and Olson 1991a, 1991b; Watts and Peet 1993). Critical constructs of RPE include:

The application of RPE is a potentially complex undertaking. A useful device to assist in application is the Kite Framework (Campbell and Olson 1991a). This framework is designed to outline the structure of these dynamic interactions through time and across scales (Figure 1). It assists NR managers in devising a strategy to describe key issues and define the fundamental causal processes underlying them.

II.B. Methodology for Assessing and Explaining Land Use and Land Cover Change
The principal methods for assessing land use change in the pilot project area are interpretation of aerial photography and analysis of MSS LANDSAT imagery.

II.B.i. Interpretation of aerial photography.
The traditional approach to change detection using aerial photography has for a long time relied on visual and descriptive methods of analysis. This approach lacks concise quantification and querying capabilities. Integrating aerial photography and GIS in change detection introduces the digital thematic mapping capability and data management of GIS that in combination with visual methods affords a quality final result.

Two air photo sets, 1973 and 1990,(click here for pictures) were received from the Institut Géographique National in Paris, with permission of the Minister of Public Works, Government of Rwanda. Both photo sets are at the scale of 1:50,000. The 1973 photos are black and white panchromatic while the 1990 imagery is black and white infrared. They represent similar seasonal surface patterns in terms of land use and vegetative cover (both were acquired during August).

A four-step process was used to derive results of the analysis:
Photo Interpretation. In both sets of photographs the average scale was 1:50,000. This scale can only approximate the actual scale at all points covered by the photographs. There were three flight tracks in each set. Changes in tip, tilt, or flight altitude above terrain height, as well as differences in terrain relief, collectively affect the overall geometry of the photos, and thus the planimetric fidelity of the final digital product.

Classification. The identification relied on the use of a topographic map (based on 1973 aerial photographs, updated with 1:20,000 airphotos taken in 1977 and 1982, and field checked in 1986) and Jennifer Olson's familiarity with the region. The visual interpretation of aerial photography depended on the basic elements: shape, size, texture, tone, and feature proximity to delineate land cover types. The interpretation was done with stereoscopic viewing using mirror stereo-scopes.(click here for pictures)

At the scale of 1:50,000 one square centimeter represents twenty five hectares. The minimum map unit size was 5mm X 5mm, a size dictated by pen width and the ability to visually distinguish landscape features. This was considered a workable resolution, but it does create a problematic level of generalization. Five millimeters square at this scale represents about 6.25 ha -- large parcels by Rwandan standards where the average farm size is 0.7 ha!. Put another way, the minimum mapping size chosen for this project is equivalent in area to nine contiguous, average-size Rwandan farms.

With this limitation, nine broad categories of land cover were recognized in the two sets of images (Table 1). These types were based on the categories used on the topographic maps of the area supplemented with additional sub-classes (wooded pasture, agricultural frontier, project pasture, and project cultivation). These additional classes were created because they are important in the analysis of the societal processes that occur on the landscape.

Photo mapping. Photomorphic regions were compiled onto a stable-base copy of the topographic map using a Zoom Transfer Scope (ZTS). The ZTS allows enlargement, rotation and differential magnification ratio in X and Y directions (i.e. an optical stretch). It is also possible to adjust the amount of light on the photo and the base map surfaces independently which allows accurate registration of the two. These qualities makes the ZTS a preferred instrument for adjusting the photo image to control points on the base map, and it facilitates interpretive overlay of the landscape pattern from the contours with the features delineated from the photography.

Integrated digital/visual. The maps of photomorphic regions for each time period were digitized in a vector format into a GIS using ESRI's ARC\INFO. The operation was done in a UNIX environment using a SUN Workstation. The GIS format enables convenient cartographic mapping. It is also easy to manipulate the database. Viewing, querying, and generating output also becomes easier and faster.

TABLE 1. LAND USE CLASSIFICATION FOR AIRPHOTO INTERPRETATION

Forest Natural forest cover. Rough, heterogeneous texture, dark tone.
Forest savanna(click here for pictures) Riverain grasslands within the natural forest. Smooth texture, light tone, within natural forest.
Planted trees (click here for pictures) Trees, usually of homogenous species (often eucalyptus or pine), planted contiguously in plots. Evenly patterned (in lines), dark tone, smoother texture than natural forest.
Tea plantation (click here for pictures) Tea bush fields and associated factory buildings. Very light tone in IR photos, light grey tone in panchromatic. Very smoothly textured. Several large rectangular buildings in center of area.
Wooded pasture Numerous scattered clumps of trees in pasture land. Light tone and relatively smoothly textured area with frequent, irregularly placed darker masses with a rough texture. Class found along natural forest boundary.
Frontier Area predominately pasture but with lines of houses and fields along valleys. Smooth texture and light color on hillsides with small buildings and associated small darker blocks. Along roadside, and between area of intense cultivation and pasture.
Project cultivation Area with radical terracing and cultivated fields planted with single crop. Patterned lines of steep terraces in area within project boundaries.
Project pasture "Improved," planted grassland with no clumps of trees for pasturing project animals. Very smooth, homogenous texture, light tone.
Cultivation(click here for pictures) Settled areas by farmers with almost continuously cropped surfaces. Very heterogeneous: mixed tones often in small block patterns, rough texture. Common single trees or short lines of trees. Some terracing.

Errors. The final output is subject to a number of human and technical (mechanical) errors that may have a cumulative impact over the various stages of the mapping process. These potential errors include:

  1. Errors of omission/commission in the photointerpretation process due to film resolution and/or contrast.

  2. Conceptual error is associated with the definition of land cover categories. The potential for these types of errors was minimized by having people familiar with the study area do the photointerpretation. Ground truthing would have improved the categorization process but this was not possible due to the security situation in Rwanda.

  3. Cartographic error: Thematic errors can be made in assigning categorical values to map features (i.e. correctly interpreted by wrongly labeled). Measurement error reflects any imprecision in the location of significant identifying features or of attributes associated with categories. These types of errors were minimized by having the two interpreters cross-check each other's work.

  4. Digitizing error. Even the most proficient operator cannot digitize the same line twice in exactly the same fashion. Several types of incorrect digital line work can result (e.g. sliver polygons, dangles, overshoots, etc.). To minimize digitizing errors registration consistency with the tics was ensured, a suitable level of generalization was employed (fuzzy tolerance and node snap distance).

II.B.ii. Interpretation of LANDSAT MSS Images
A Landsat-1 MSS scene (March 12, 1975) and a Landsat-5 MSS (July 3, 1986) were evaluated for use in this study.(click here for pictures) The imagery was acquired from EOSAT, Inc. (Lanham, Maryland, USA). Subscenes covering the northern part of Nyungwe Forest were extracted from both images. These subsets are considerable larger than the test area for which aerial photography was available (5,215 km2 versus 763 km2).

For a variety of reasons, the MSS analyses could not be used effectively to assess changes in land cover. These included:
Seasonal differences in imagery. The March scene (1975) represents the wet season in this part of Africa; the July image (1986) captured the dry- season appearance of the landscape. These differences were problematic regarding change detection, but were necessary due to the availability of relatively cloud-free imagery across a long time interval.

Insufficient data. The 1986 imagery contained corrupted and unusable band 2 data (red reflectance). As a result, despite attempts to compensate by using only three bands in both images - bands 1, 3, and 4, it was obvious from the evaluations that the classifications for either scene were seriously confused for large portions of the landscape. The training statistics for many of the classes were unacceptably heterogeneous, causing serious errors of commission and omission (based on visual scrutiny, since ground truth data were unavailable).

The chief cause of the problems with the digital classification approaches in this study, both supervised and unsupervised, was the high-frequency illumination differences caused by the dissected topography. These could have been corrected using digital elevation data. One of the more disheartening aspects of this study was that a DEM exists for the whole of Rwanda, but these data were not made available to this project.

II.C. Socio-Economic Analysis
A variety of methods and sources of data have been used in the compilation of information regarding the population on the edge of the Nyungwe Forest. These include household surveys, group interviews with farmers, interviews of officials, and library research of historical documents, government reports and scientific articles.

II.C.i. Review of literature
The review of literature is important in providing an historical context and information about a variety of domains - policy, society, religion, agriculture, economic trends, case studies etc.

II.C.ii. Government and project reports
A decade ago Rwanda was a data-poor country, yet today it has one of the most comprehensive data sets in sub-Saharan Africa. This change is a consequence of a number of important initiatives to improve data availability taken by the Government of Rwanda with the support of donor agencies, such as USAID and Belgian Technical Cooperation.

The data base includes population censuses, aerial photography, and rainfall statistics, agricultural surveys begun in 1983, and the soil map released in 1993. These data gathering activities are complemented by government reports from various ministries, and by the data bases and reports of the many development projects that have been initiated in the country.

Most of this information represents national issues and trends, and the smallest geographical unit is usually the commune. As a result, local issues cannot be addressed with the same wealth of information as national ones. However, the national context for local issues can be clearly defined.

II.C.iii. Household surveys (click here for pictures)
Some information, presented in a tabular format, was derived from statistical analyses of household survey results. The data, from the 1991 Survey of Agroforestry, are from a nationwide stratified random sample of 1240 farm households (operating 6,464 fields) by the Division of Agricultural Statistics of the Ministry of Agriculture. Interviews with heads of households and/or their spouses were conducted in June 1991. The survey included a field-level component (of variables such as changing soil fertility, number of trees and tree species, etc.) and a household-level component (e.g., changing fuelwood availability, distance to fuelwood, reason trees were planted).

288 households in the three prefectures surrounding the Nyungwe Forest, Cyangugu, Kibuye and Gikongoro, were available from the Agroforestry Survey. This survey was designed to be statistically valid at the national and at the prefectoral level. The results presented in this report of three prefectures should, therefore, not be rigidly interpreted as statistically valid but, in the absence of our ability to conduct a larger sample, they may provide a general indication of how responses vary between areas.

II.C.iv. Informal surveys
An essential activity in research that examines society-environment interaction is verification of the interpretations based on literature reviews, official reports and statistics, and survey data - the equivalent of ground-truthing in remote sensing analyses. Procedures for this include group discussions and interviews with key informants. It should be emphasized that in some cases, particularly where costs are to be minimized, these activities can, with careful planning and implementation, replace formal surveys and yield appropriate and useful information.

II.C.v. Group interviews(click here for pictures)
In many African rural societies formal group meetings are a regular feature of social, religious and political life. Such meetings have socially defined rules of order, and in many cases certain segments of the society may be excluded or have only peripheral roles.

Group meetings have the advantage that they can be organized to focus on specific issues, such as the environment, and in the course of discussion the group acts as its own arbitrator of the validity of information and of its interpretation. Experience has shown that they can provide extraordinary insights to issues, insights that go far beyond what might be achieved from intensive survey work alone (Campbell 1984). Group meetings can also be very difficult experiences and can provide misleading information. Political circumstances, and failure to respect local custom as to format, attendance and participation can prevent a productive meeting (Campbell 1987). Further, care has to be taken to assess the biases inherent in these situations such as those arising from age, gender, or socio-economic prejudice in the participant group.

Much of the information presented in this report on the interaction of people with natural resources was obtained from discussions with villagers in the study area (Olson 1994c). The meetings were structured around the findings of household surveys that involved questionnaire interviews at the household level. The discussions at the group meetings went far beyond the specific questions and provided " a broad historical and especially political context to supplement the narrower answers given by individuals concerning their own fields and farms." (Olson 1994c: 69).

II.C.vi. Key informants
While the group meetings can provide details on the broad context for local circumstances and practices, there are frequently particular individuals who for whatever reason, have acquired significant knowledge about specific issues. These individuals come from a variety of segments of society , including farmers, bureaucrats, elders, priests, and local "historians." What sets them apart as key informants is that they are recognized by others in their community as being particularly knowledgeable about the area, or facets of it.

III. RESULTS
Land use change has taken place in the context of the farming systems adjacent to the Nyungwe Forest. The Nyungwe Forest forms a physical barrier between two separate farming systems on either side, the Lake Kivu Shore and the Gikongoro farming systems. While the entire area falls within one agro-climatic zone, the Zaire-Nile Ridge, and has generally similar conditions for crop growth (Delepierre 1975), historical cultural, social and economic legacies continue to affect crop and animal production patterns.

The differences between the farming systems of Lake Kivu Shore and Gikongoro are reflected perhaps best in the relatively minor crops and animals the two regions produce. Gikongoro, for example, is the site with the highest concentration of pig production in the country whereas Lake Kivu Shore has very few pigs. Similarly, Gikongoro produces much more eleusine (a grain), sheep, sweet potatoes and wheat than Lake Kivu Shore. On the other hand, Lake Kivu Shore produces more coffee, goats, maize, bananas and soybeans than Gikongoro. The differentiation of the farming systems on the eastern and western sides of Nyungwe Forest is also reflected in levels of development and income, and tree use and tree planting are all distributed along a east/west divide.

III.A. Land Cover Changes 1973-1990 (click here for pictures)
The results discussed in this section concentrate on the land cover changes detected using air photography for 1973 and air photography for 1990, supplemented in areas of unmatched coverage with the 1973 photography by information from a 1:50,000 scale topographical map with land cover indicated for 1986. Significant changes are the emergence of land uses associated with national policies, the expansion of the cultivated area, and the disappearance of the wooded pasture and the agricultural frontier.

Policy-related land uses: (click here for pictures) the green belt, designed to act as a buffer between the natural forest and smallholder cultivation; tea plantations, and demonstration projects for improved cultivation, and pasture. The green belt was planted on land previously occupied by wooded pasture, natural forest, and pasture. It was the most important cause of decline in the area of natural forest, replacing almost 16 km2 of natural forest in the area studied.

Cultivation: (click here for pictures) cultivation did expand, from 260 km2 in 1973 to over 370 km2 in 1990, mainly at the expense of pasture and the agricultural frontier. This indicates the success of the green belt in acting as a buffer between agricultural activities and the natural forest. No incursion of cultivation into the natural forest is shown in the area analyzed with air photography. This finding may not be surprising given that the analysis was conducted on the eastern edge of the Nyungwe Forest, in communes that have been undergoing out-migration. We cannot be as confident about cutting for cultivation on the western side of the forest as population increase and in-migration have been occurring in the communes of the area.

The results of the study indicate that the area of the Nyungwe Forest has declined, but the overall area of forest, natural and planted, has increased by just over 6 percent of the total air photo study area of 763 km2. The decline in the area of the natural forest is not a consequence of the expansion of smallholder agriculture as would be indicated by the concerns of national policy-makers and donor agencies, but rather it is the result of government projects along the edge of the forest designed to act as a buffer against the encroachment of agriculture. The result has been effective in that cultivators have not been able to cut the forest to obtain farm land.

In terms of conserving biodiversity in the study area, two opposing processes have been identified. First, the area of the natural forest has declined, but we have no information on the impact of the reduction of the area upon the biodiversity of species in the forest. Second, over the same 20 year period, the biodiversity of the agricultural landscape adjacent to the Forest has increased as farmers have selected a variety of crops and have planted a large number of endogenous tree species on their lands. The minimum mapping size chosen for the airphoto analysis precluded the mapping of most of the trees that have been planted throughout the agricultural landscape. In consequence, our estimate of the total amount of arboreal vegetation in the study area is an underestimate. Recent research in the area confirms the frequency of tree planting by individual farmers (den Biggelaar 1994; Olson 1994c).

In sum, it is likely that in the area of study, the total forested area and the biodiversity of the forest-agriculture landscape have increased. The finding that agricultural activities have increased on-farm biodiversity and have not contributed to a reduction in the area of natural forest is counter-intuitive.

III.B. Explanation of Land Cover Change 1973-1990
The people of the area have been actively attempting to maintain and improve their quality of life in the face of difficult circumstances arising from the quality of the resource base, demographic pressures, and the impact of government policies. These can be discussed in terms of four major, interacting themes: pressure on resources, the evolving crop/animal/tree triad over time, the use of trees (on versus off-farm), and migration.

III.B.i. Pressure on resources (click here for pictures)
Over the past 20 years or so each of the farming systems has had to contend with a number of processes that have placed considerable pressure on the resource base. These include population growth, a loss in communal lands, low prices for agricultural products, and limited off-farm economic opportunities. These processes are not independent of each other but rather interact to create conditions of pressure on resources.

III.B.i.a. Population pressure. Rapid population growth is most commonly mentioned cause of pressure on resources as increasing numbers of people cultivate a limited amount of land. This has been an important factor in Rwanda which at its peak period of growth in the late 1970's experienced a growth rate of around 3.7 percent or a doubling of the population every 19 years (MINIPLAN 1982).

The population started to grow rapidly after World War II with the introduction of new crops and new agricultural techniques, a decline in diseases due to improved medical care and hygiene, and the abandonment of traditional forms of child spacing with the conversion to Catholicism (ONAPO 1990). Population densities rose from 127 km2 in 1948 to 327 km2 in 1978 and finally to 495 km2 in 1991 in Cyangugu. In Gikongoro, the rise was less dramatic but nevertheless very rapid with a growth from 105 km2 in 1948, through 242 km2 in 1978, to 302 km2 in 1991 (Prioul and Sirven 1981; MINIPLAN 1982; MINIPLAN 1992). The rise in population densities has resulted in an increase in cultivation as pasture, valley and forest land was converted to crops. This resulted in a loss of grass for pasture, fuelwood, medicinal plants, grasses for making baskets, and to use as green manure, etc., from communal land that had been maintained for hundreds of years for those purposes. Finally, with a cultural system in which each son inherits an equal amount of land from his father, each generation has seen a shrinking of farm sizes in areas where supplemental land to clear or buy is difficult to obtain.

III.B.i.b. Loss of communal lands. Before independence in 1962, most of the population was confined to the foothills of the western highlands by a tightly-controlled, hierarchical political system in which the Hutu farmers had little power. Access to land was controlled by the cattle-owning Tutsi (Newbury 1988). The savanna East and entire communes in the West (e.g., Musange in Gikongoro) were reserved as pasture. After independence, when a power reversal resulted in the Hutu coming to power, the East was opened to settlement by farmers resulting in waves of migration from the densely populated west.

The independent government continued to control access to land. In the area surrounding Nyungwe Forest land was allotted to parastatal tea plantations and to economic development cum forest preservation, projects financed by foreign governments. The appropriated land had been used in an extensive manner by local farmers for seasonal pasture, fuelwood and/or crop production. In contrast, the plantations and development projects were intended to generate foreign exchange for the government.

Local farmers have gained few direct benefits from the plantations or the projects. Research has concluded that on-farm tea production is less profitable than food crop production, and that households with a member working on the plantation have actually less food to eat than other households (Laure 1986; von Braun et al 1991). Farmers interviewed near the Crête Zaïre-Nil (CZN) project in northwest Gikongoro are extremely bitter at the project and the government for having appropriated their land without recompense (actually, 2,624 families were relocated but to extremely marginal land) and providing jobs and other benefits of the project to people from outside the area.

III.B.i.c. Low prices for agricultural produce. A third factor in the increased pressure on land resources has been the low revenues gained by farmers, intensifying their poverty and desperation (Ben Chabaane 1992; World Bank 1993b). Unfavorable terms of trade, a limited national market for agricultural products and declining coffee prices on the international market have resulted in low prices at the farm gate. This is particularly the case in the Southwest which is less tied to the nation-wide commercial network and produces less of the high-value crops. Some areas with highly acidic soils, as in highland Kibuye and Gikongoro, are also suffering from severe soil degradation and declining productivity, further increasing their poverty (Olson 1994a). Lack of access to productive resources (Blaikie and Brookfield 1987), including animals for manure, improved seeds, and fertilizers, is also felt differentially between households within regions depending upon their ability to afford the inputs.

III.B.i.d. Limited non- and off-farm opportunities. The low farm revenues and declining availability of farm land have led to agriculture alone being unable to support the rural population in much of the West. This has increased the importance of the role of non- or off-farm sources of income in rural areas. Again, the Southwest had fewer opportunities than other areas as the former President Habyarimana and the people he had selected for important governmental posts concentrated governmental funds in their home region (Newbury 1992). Within the Southwest, Cyangugu was somewhat favored with trade opportunities with Zaire and Burundi. Farmers in Gikongoro and Kibuye, however, were relatively isolated from any trade network or other opportunities outside of subsistence agriculture.

This concentration of "proximate causes" of pressure on resources in the Southwest relative to the rest of the country are reflected in this area having the lowest incomes and experiencing some of the worst symptoms of stress such as poor child nutrition and soil degradation (Von Braun et al. 1991; Tardif-Douglin et al. 1992; Olson 1994a).

III.B.ii. The Evolving Crop/Animal/Tree Triad Over Time
Long-held resentments of the Tutsi and the Belgians led to the eruption of the Social Revolution and Independence in 1962. The Tutsi lost to the Hutu all political power and control over access to land and other resources (Lemarchand 1970). The pent-up demand for cropping land was released as land previously reserved for pasture became available to farmers. Two immediate impacts of the revolution were 1) reserved pasture land locally in the West and the entire East of Rwanda was opened up to farmers for settlement, and 2) colonial policies towards agriculture were reconsidered by the newly independent government. These political events, as well as demographic and economic changes, led to a major shift in the relative proportion of land devoted to crops, trees and animals.

One policy initiated very early by Belgian colonialists, and then continued in the 1980s by the post-independence government, was exotic tree plantings with umaganda (forced communal labor) on government woodlots and along roads. The eucalyptus, cypress and pines were to produce construction materials for the government buildings and churches and to reduce erosion from the hilltops. In the early 1980's the government established tree planting on arbor day. Expatriate agro-forestry projects and governmental programs created tree nurseries and actively encouraged planting of trees, such as grevillea and eucalyptus, as part of an anti-erosion campaign. The results include an increase in availability of seedlings and knowledge of tree planting, more trees in government woodlots, and more trees on private farms. The governmental extension efforts were coincident with an increase in the demand for tree products (fruit, fodder, medicines, wood etc.) as the population increased and as forest remnants were converted to crops.

In these regions, therefore, there has been a large increase in planted trees on large and small farms as the few wild trees that had been managed in forest remnants disappeared. Farmers had previously planted fruit trees and euphorbia around their houses, but now trees are also planted on field boundaries and elsewhere on the farm (den Biggelaar 1994). Trees have become similar to other crops in terms of their economic importance, and indeed have replaced cattle as a source of long-term savings. Trees are carefully managed and provide a constant source of food, fodder, medicine and fuel, and small to medium-sized poles which can be sold (den Biggelaar 1994). Trees play a different role in the household economy from that of crops, however, since they are owned by men and usually provide an infrequent if larger sum reserved for major purchases or to repay debts.

III.B.iii Trees: On versus Off-Farm
The most common use of Nyungwe Forest resources by neighboring farmers was probably the collection of wood products, especially fuelwood. There are many other purposes for which farmers use trees, including for medicines, yeast for beer production, stakes for climbing beans, etc. Some of these uses require certain species of trees, whereas the requirements for fuelwood, the primary use of trees, are more flexible. The importance of tree products for farmers' daily life is thus critical. Changes in availability of tree products are, therefore, quickly and intensely experienced.

The results of the study demonstrate that farmers near the forest had, in the past, collected fuelwood from Nyungwe Forest and grazed their animals on land surrounding the Forest. In the past few years, they have lost their access to both the forest trees and the pasture land due to appropriation of land by greenbelt projects. Those farmers in Cyangugu near the forest have compensated for the loss by planting trees on-farm and have therefore somewhat reduced the impact of the loss of their access to the forest. The farmers near the forest in Muko, however, did not benefit from governmental or expatriate programs encouraging tree planting and they have not yet adapted their farming system to planting trees on-farm. On farms farther from Nyungwe Forest, sources of tree products became scarce as unclaimed communal land was cleared and converted to crops. However, farmers planted sufficient additional trees on their farms to the point where they were able to meet their fuelwood needs on-farm.

In the Southwest, the land under grass has declined, that under trees has somewhat increased, and the land under crops has enormously increased. The future of this trend is uncertain since farmers find that there is no longer any uncleared land to be put under crops. Meanwhile, the value of animals remains high since they produce the precious manure for crops. Growing fodder (grass and tree leaves) on-farm is beginning to be a necessary part of the agricultural system. The necessity of planting trees and grasses on-farm has changed the landscape. Grasses had been found in communal or private pasture lands away from the intensely cropped fields surrounding the house, and the rare trees had been in valleys or hilltops on communal lands. Now grasses and trees are planted on the edges of fields as part of the cropping system. Another recent trend is that farmers with extremely degraded cropped land are being forced to convert the land to pasture or woodlots, in a sense completing the cycle. This continuing process of agricultural intensification, coupled with the ecological process of land degradation, may lead in the future to somewhat less land under crops and more under grass and trees.

III.B.iv. Migration
The reality of limited additional agricultural intensification opportunities and rare non-agricultural employment in the Southwest has led to adoption of another strategy, that of migration. Out-migration to the East began after independence in 1962 and by 1980 the population had seemed relatively evenly distributed throughout the country. The results of the 1991 census of the Rwandan population indicate, however, that migration as a response to resource constraints and to perceived regional differences in economic opportunities remained important in the period 1978 to 1991. The possibility of emigration by farmers seeking land had become increasingly impossible due to the unwillingness, and at time hostility, of neighboring governments and farmers of Uganda, Zaire, Burundi and Tanzania to consign their valuable and/or scarce land to foreigners (Sirven 1984; Bart 1993).

By the early 1990s, the opportunities at the new destination areas in Rwanda are much poorer compared to what earlier generations had enjoyed. The only zones of new land that migrants could claim were in the dry, rocky far East, in marsh land that required draining, or alongside forest reserves. The sole town receiving a large number of migrants was the capital city, Kigali-Ville, but the absence of good opportunities for those less than well-educated meant that rural areas continued to attract substantially more permanent migrants than cities.

The extreme concentration of rural in-migration to a few communes in the far East reflects what many authors have written and what farmers have said, that opportunities for finding new land in Rwanda were extremely limited (Uwizeyimana 1991; Guichaoua 1989; Bart 1993). It is interesting to note that the communes alongside Akagera Park are almost all receiving in-migrants. Whether the new settlers are finding land outside the Park boundaries, as previous migrants had done, or moved into the Park and hunting reserve, is unknown.

A few rural communes in the Western Highlands are also experiencing a small degree of in-migration. These communes are located alongside forests, indicating that in-coming farmers are probably clearing lands next to the forests or inside the official forest reserve boundaries. This is visible along the Parc des Volcans in the Northwest, the former Gishwati Forest Reserve and the western edge of the Nyungwe Forest Reserve in the Southwest. Approximately 8,600 net migrants settled along the western edge of the Nyungwe, mostly south of the road (Karengera and Bugarama communes). The in-coming farmers to the eastern edge of the forest were probably short-distance migrants, coming from the surrounding communes, and perhaps some from Gikongoro. Much of the land along the eastern edge of Nyungwe had been expropriated for parastatal projects and this, along with the poor soils and widely-known poverty of local farmers, apparently kept people from migrating to this land.

These movements towards the parks were undetected in earlier migration studies (Olson 1990; Cambrezy 1984) either because they are a new phenomenon or because they were not detected in the broader inter-prefectoral movements studied. One indication that they are a new phenomenon is that the 1978 to 1991 inter-prefectoral movements show a new stream of migrants moving from Gikongoro to Cyangugu, probably to those communes along the western edge of Nyungwe Forest receiving in-migrants.

This extension of the historic West-to-East rural-rural flows seems rather final, literally the end of the road. The only unsettled land remaining in Rwanda is found in the 15 percent of country in parks and forest reserves, land set aside initially because it appeared to have little agronomic potential. It is in this context that concern over the future sustainability of the Nyungwe Forest arises. Past agricultural activities had not significantly depleted the forest, either as a direct result of farmers' decisions and/or due to the protection afforded by the buffer zone of projects. Most farmers sought alternatives in activities that did not threaten the forest. However, as non-farm opportunities have been insufficient to meet the farmers' needs, so trends towards cultivation of the edges of the parks and reserves have increased. These trends have been completely disrupted by the dramatic impact of the 1994 civil war on the numbers and distribution of the population. The impact on the forests and reserves is as yet unknown. However, the lesson that non-farm opportunities are required to reduce land degradation and the threat to the country's parks and reserves should not be ignored by those planning Rwanda's future development.

III.C. Summary
This study has focused on an identification and explanation of land cover changes that have occurred over the past 20 years in a small area of South West Rwanda, including a portion of the Nyungwe National Forest. Considerable concern has been expressed in government and conservationist circles over the need to protect the biodiversity of the Forest in the face of a presumed threat from the expansion of agriculture, driven by a rapidly growing population in adjacent areas.

The results of the study indicate that the area of the Nyungwe Forest has declined, but the overall area of forest, natural and planted, has increased by just over 6 percent of the total air photo study area of 763 km2. The decline in the area of the natural forest is not a consequence of the expansion of smallholder agriculture as would be indicated by the concerns of national policy-makers and donor agencies, but rather it is the result of policies designed to protect the forest through the establishment of government projects along the edge of the forest to act as a buffer against the encroachment of agriculture. The result has been effective in that cultivators have not been able to cut the forest to obtain farm land. They have, however, virtually replaced pasture with cultivation and extended their activities up to the buffer zone.

In terms of conserving biodiversity in the study area, two opposing processes have been identified. First, the area of the natural forest has declined, but we have no information on the impact of the reduction of the area upon the biodiversity of species in the forest. Second, over the same 20 year period, the biodiversity of the agricultural landscape adjacent to the Forest has increased as farmers have selected a variety of crops and have planted a large number of endogenous tree species on their lands. The minimum mapping size chosen for the airphoto analysis precluded the mapping of most of the trees that have been planted throughout the agricultural landscape. In consequence, our estimate of the total amount of arboreal vegetation in the study area is an underestimate. Recent research in the area confirms the frequency of tree planting by individual farmers (den Biggelaar 1994; Olson 1994c).

In sum, it is likely that in the area of study, the total forested area and the biodiversity of the forest-agriculture landscape have increased. The finding that agricultural activities have increased on-farm biodiversity and have not contributed to a reduction in the area of natural forest is counter-intuitive.

This outcome of the study illustrates the importance of defining the concept of biodiversity carefully so that appropriate and effective policies can be enacted to maintain it. In the study area biodiversity is represented in a contiguous protected area with significant flora and fauna - the Nyungwe Forest, and also in the trees and crops selected by farmers. Smallholder cultivation and forest protection are not necessarily in conflict, and in this area apparently combine to promote vegetational biodiversity in the landscape.

IV. METHODOLOGICAL IMPLICATIONS OF THE STUDY
This Pilot Project was designed to provide insights to appropriate ways of examining such interactions through a preliminary analysis of society-forest interactions in communities adjacent to the Nyungwe Forest Reserve in Rwanda. The study examined the changes in land use and land cover in the areas of the forest and adjacent to it and looked for explanations for these changes in the processes of national and local development.

The study has concluded that the following are critical components of an effective methodology:

IV.A. A Conceptual Framework.
A conceptual framework is needed that leads the researcher to examine the complex web of linkages that determine the interactions between society and natural resource management. These interactions occur between different scales (global, national, local) over time and space and so researchers have to explore the evolution of these interactions over time both at the local level, and between rural communities and national, and even global, forces.

The conceptual framework played a signal role in defining the approach to analysis of land use changes related to biodiversity conservation in the study area. It also clarified the relevant questions for assessing the relationship between societal processes and natural resources management over time. Finally, the framework established the structure for the presentation of the findings. It is an essential component of the methodology used by the project.

IV.B. A Historical Perspective.
A prevailing perspective among government officials, researchers and donor groups, is that rapid deforestation in Rwanda is a relatively recent and active process resulting from rapid population growth. The results of work by Olson for this study, and for her dissertation (Olson 1994c), has demonstrated that the situation is more complex:

IV.C. Information Needs.
The critical questions derived from the application of the conceptual framework defined the information needs of the study. Two broad categories of information were required: information on the extent, location and timing of changes in land cover; and information on the processes that contributed to these changes.

The availability of official reports is limited by the fact that they are not centrally archived, nor are they usually shared between government or other institutions. There is an urgent need to explore ways of managing such information in ways which permit it to be accessed more easily and analyzed in a more integrated manner (Campbell et al. 1994a; 1994b).The availability of official reports is limited by the fact that they are not centrally archived, nor are they usually shared between government or other institutions. There is an urgent need to explore ways of managing such information in ways which permit it to be accessed more easily and analyzed in a more integrated manner (Campbell et al. 1994a; 1994b).

IV.D. Analytical Methods

IV.D.i. Remote Sensing.
Changes in land cover were determined using aerial photography acquired in1973 and 1990, supplemented by the 1986 topographic map. Several observations concerning the appropriateness of these data sources can be made:

  1. For countries like Rwanda where the units of cultural land cover occur at such high spatial frequencies, the use of aerial photography at scales of 1:50,000 or larger is judged to be necessary to correctly interpret detailed land cover changes in terms of social processes.

  2. MSS is adequate to interpret level-one land cover types which vary at low spatial frequencies (measured in hundreds of meters) if the terrain is reasonably undissected (broad interfluves and valley bottoms more than 300-500 meters across).

  3. Multi-temporal TM imagery is probably required from each of two time periods in order to assess land cover change at level-two or higher.

  4. Although difficult to obtain, historical aerial photography does exist for many countries in Africa. This valuable data source should be used more often both as a source of historical ground truth for achieved satellite imagery and as primary data in land cover change analyses as demonstrated in this study.

IV.D.ii. Interpretation of land Use/Cover Change.
The remotely sensed data permit estimates of the extent and the location of major land cover change. The critical question for biodiversity conservation policy is how do we interpret these changes? Effective policy requires an understanding of the fundamental causes of change so that it can address these causes and not merely symptoms.

The conceptual framework guides the exploration of driving forces through interpretation and explanation of societal processes that lie behind the mapped changes in land cover. This requires the integration of information from a variety of sources including:

IV.D.ii.a. Historical Records.
Documentation of past processes that contributed to recent and ongoing changes was obtained from colonial documents, books and reports of scientific research. Historical records, photographs and reports of early European travelers in the region caused contemporary assumptions about past land cover to be questioned. Once these assumptions were shown to be suspect, critical examination of the processes lying behind land cover change became essential.

IV.D.ii.b. Policy Documents and Official Reports.
These sources of information were found to be available in Rwanda, but were scattered. The availability of official reports is limited by the fact that they are not centrally archived, nor are they usually shared between government or other institutions. There is an urgent need to explore ways of managing such information in ways which permit it to be accessed more easily and analyzed in a more integrated manner (Campbell et al. 1994a; 1994b).

Policy documents assist in the understanding of official initiatives that have an impact upon conservation issues. Some, such as the NEAP, relate directly, while others, such as economic policy documents, have to use interpreted to reflect the impact of economic initiatives upon resource management.

IV.D.ii.c. Interviews.
The project was fortunate to have been able to draw on a rich body of recent information gathered at the household, and from individuals and groups. The outbreak of civil strife in Rwanda in April 1994 precluded specific survey work for this project.

The interviews were essential to the interpretation of the causes behind changing patterns of land cover. In the absence of interviews, official data would have been relied upon. Yet such data can be misleading if it is not used in an analytical context that accurately reflects the conditions under study.

A number of interview strategies was identified. Sample household surveys provide the most detailed information, but are expensive and reflect a significant institutional infrastructure. At the other extreme, Rapid Rural Appraisal may provide significant information if conducted carefully. Group interviews and discussions with key informants were important to this study. While all these can yield useful data, the basis of each needs to be considered explicitly. The findings of this project would suggest that multiple approaches, that provide opportunities for cross-checking and corroborating information, are desirable. Further, the information they provide has to be evaluated and analyzed within a context informed by the historical record and policy framework.

IV.E. Conclusion
This pilot study has demonstrated that understanding the complex interactions between society and the environment as a basis for effective NRM policy, including conservation of biodiversity, requires an approach that:


BIBLIOGRAPHY

Barabwiliza, Runyinya. 1980. Impact de l'Homme sur les Formations Forestière des zones Marginales au Rwanda. Bulletin Agricole du Rwanda 13: 14-23.

Bart, François. 1993. Montagnes d'Afrique Terres Paysannes: Le Cas de Rwanda. Centre d'Etudes de Géographie Tropicale, Espaces Tropicaux No. 7. Bordeaux, France: Presse Universitaire de Bordeaux.

Ben Chaabane, Hichem, Prosper Cyiza and Augustin Rushengura. 1991. L'Indice des Priorités Communales: Un Indice Synthetique du Taux de Développement des Communes Rwandaises. Kigali: MINIPLAN, Rép. Rwandaise.

Ben Chaabane, Hichem and Prosper Cyiza. 1992. Méthodologie d'Elaboration de Données sur le Revenu Rural des Communes Rwandaises. Kigali: MINIPLAN Direction de la Planification Division des Stratégies de Développement Communal et Régional.

Berry, Eileen, Len Berry, D.C. Major, and David J. Campbell 1982. "Regional reconnaissance of Rwanda, Burundi, Kivu Province of Zaire." Final Report to USAID. Program for International Development, Clark University, 1982. Two volumes.

Blaikie, Piers M. and Harold P. Brookfield. 1987. Land Degradation and Society. New York: Methuen.

Brown, Daniel G. and Daniel A. Griffith. 1996. Spatial Statistics and GIS Applied to Internal Migration in Rwanda, Central Africa. In Handbook on Spatial Statistics. Boca Raton, Florida: CRC Press, Inc.

Cambrezy, Luc. 1984. Le Surpeuplement en Question: Organisation Spatiale et Ecologie des Migrations au Rwanda. Paris: ORSTOM.

Campbell, David J. 1984. "Community participation in field research: an example from Kenya Maasailand." Occasional Paper, 4. Center for Advanced Study of International Development, Michigan State University.

_______ 1987. "Participation of a community in social science research: a case study from Kenya Maasailand." Human Organization 46,2:160­167.

Campbell, David J. 1994. Environmental Stress in Rwanda No. 1: Preliminary Analysis. Rwanda Society-Environment Project Working Paper 4. Department of Geography, Michigan State University.

Campbell, David J., Len Berry and Noberto Fernandez. 1994. "Institutional Structures to Facilitate Data Collection, Management and Analysis for NRM in Sub­Saharan Africa." Rwanda Society-Environment Project, Working Paper 9.

Campbell, David J., Len Berry, Jennifer M. Olson and Daniel G. Brown. Basic Data Systems for Natural Resource Management: Findings of the Rwanda Case Study. Final Report of the Michigan State University Rwanda Society-Environment Project November 1994.

Campbell, David J. and Jennifer M. Olson. 1991a. "Framework for environment and development: the Kite." CASID Occasional Paper, 10. CASID, MSU.

Campbell, David J. and Jennifer M. Olson. 1991b. "Flying the Kite in Kajiado District, Kenya." Centennial Review, XXXV,1:295-314

Campbell, David J., Jennifer M. Olson and Len Berry. 1993. Population Pressure, Agricultural Productivity and Land Degradation in Rwanda: An Agenda for Collaborative Training, Research and Analysis. Rwanda Society-Environment Project, Working Paper, 1.

Clay, Daniel C., Jean Kayitsinga, and Théobald Kampayana. 1989. L'Emploi en Dehors du Ménage au Rwanda. ASPAP/DAI Report No. 74. Kigali: Service des Enquêtes et des Statistiques Agricoles, MINAGRI, Rép. Rwandaise.

CPR (Carte Pedologique du Rwanda). 1992. Carte d'Aptitude des Sols du Rwanda. Kigali: MINAGRI and Coopération Technique Belge.

Delepierre, G. 1975. Les Régions Agricoles du Rwanda. Bulletin Agricole du Rwanda 8 (4): 216-225.

den Biggelaar, Christoffel J. 1994. Farmer Knowledge and Experimentation with Trees and Tree Cultivation in Agroforestry Systems in Rwanda. Ph.D. thesis, Michigan State University.

DGF (Direction Général des Forêts). 1991. "Rapport de mission sylvopastoralisme." Letter Number 287/DGF/09.02 of July 17, 1991 from the DGF to the Ministère de l' Agriculture, de l'Elevage et des Forêts.

Falloux, François and Lee M. Talbot, 1993. Crisis and Opportunity: Environment and Development in Africa. London: Earthscan.

Frumhoff, Peter C. and Christine L. Bergmark. 1993. Conserving Rwanda's Biological Diversity: An Assessment of Priorities and Recommendations for USAID Assistance. Draft. Kigali: USAID.

Gahamanyi, Anastase, Venant Rutunga, Gaëtan Sibomana, Runyinya Barabwiliza, Mamadou Djimde, Michel Baumer and Dirk Hoekstra. 1988. Potentiel Agroforesterie dans les Systèmes d'Utilisation des Sols des Hautes Terres d'Afrique de l'Est à Régime Pluviométrique Bimodal, Rwanda, No. 1. AFRENA Report. Nairobi, Kenya: International Council for Research in Agroforestry.

Gatali, Felix. 1992. Personal communication. Veterinarian, Projet Crête Zaïre-Nil.

Gatera, Frédéric. 1980. Accroissement Démographique et Déforestation au Rwanda. Bulletin Agricole du Rwanda 13: 28-30.

Gourou, Pierre. 1953. La Densité de la Population au Ruanda-Urundi: Esquisse d'une Etude Géographique. (Mem. no. 8, XXI-6). Brussels: Institut Royal Colonial Belge.

Grigg, David. 1980. Migration and Overpopulation. In The Geographical Impact of Migration, ed. Paul White and Robert Woods, 60-83. London: Longman Group Limited.

Guichaoua, André. 1989. Destins Paysans et Politiques Agraires en Afrique Centrale Tome 1: L'Ordre Paysan des Hautes Terres Centrales du Burundi et du Rwanda. Paris, France: Editions L'Harmattan.

Johanssen, P.F. 1915. Ruanda: Kleine Anfänge - Grosse Aufgaben. Bieleseld: Berlagshandlung der Anstalt Bethel.

Kristensen, Kurt and Shaban Turikunkiko. 1992. Report on Gold Mining Activities in the Munronzi Watershed, Nyungwe Forest Rwanda. Kigali: Natural Resources Management Project, USAID/DAI.

Lemarchand, Réné. 1970. Rwanda and Burundi. London: Pall Mall Press.

Lewis, Dale and Nick Carter (eds.), 1993. Voices From Africa: Local Perspectives on Conservation. Washington, D.C.: World Wildlife Fund.

Loupias. 1903. Tradition et Légend des Batutsi. Anthropos III, 1: 1-16.

Maquet, Jacques J. 1954. Le Système des Relations Sociales dans le Ruanda Ancien. Annales série no. 8. Tervuren: Musée Royal du Congo Belge.

MINAGRI (Ministère de l'Agriculture, de l'Elevage et des Forêts). 1989. Production Agricole en 1987: Bilan d'Autosuffisance Alimentaire par Commune et par Habitant. Kigali, Rwanda: Rép. Rwandaise.

________. 1990. Rapport Annuel 1990, Projet Crête Zaïre-Nil. Gatare, Rwanda: MINAGRI.

MINIPLAN (Ministère du Plan). 1982. Recensement Général de la Population et de l'Habitat 1978. Kigali, Rwanda: Rép. Rwandaise.

________. 1992. Recensement Général de la Population et de l'Habitat au 15 août 1991, Résultats Provisoires. Kigali, Rwanda: Rép. Rwandaise.

MIRENAMICA. 1985 (?). Carte Administrative et Routière. MINITRAP Rép. Rwandaise.

Muko. 1992. Transcription of recorded group interview of farmers in Muko Commune, Gikongoro. September.

Mutabaruka Célestin. 1992. Personal Communication. Director, Projet Crête Zaïre-Nil.

Newbury, Catherine. 1988. The Cohesion of Oppression: Clientship and Ethnicity in Rwanda: 1860-1960. New York: Columbia University.

________. 1992. Rwanda: Recent Debates Over Governance and Rural Development. In Governance and Politics in Africa, eds. Goran Hyden and Michael Bratton. Boulder: L. Rienner.

Nezehose Jean Bosco. 1990. Agriculture Rwandaise: Problématique et Perspectives. Kigali: INADES Formation.

Nzisabira, J. 1989. Accumulation du Peuplement Rural et Ajustements Structurels du Système d'Utilisation du Sol au Rwanda Depuis 1945. Bulletin Agricole du Rwanda 2:117-127.

Offutt, Katherine, Amy Vedder, Anne-Marie Gillesberg, Robert Clausen, Marc Matthusens and Dean Gibson. 1990. Nyungwe Forest Reserve. Cyangugu, Rwanda: The Nyungwe Forest Conservation Project.

Olson, Jennifer M. 1990. The Impact of Socioeconomic Factors on Migration Patterns in Rwanda. M.A. thesis. Geography Department, Michigan State University.

________. 1994a. Farming Systems of Rwanda: Echoes of Historic Divisions Reflected in Current Land Use. Rwanda Society-Environment Project Working Paper 2. Department of Geography, Michigan State University.

________. 1994b. Demographic Responses to Resource Constraints. Rwanda Society-Environment Project Working Paper 7. Department of Geography, Michigan State University.

________. 1994c. Farmer Responses to Land Degradation in Gikongoro, Rwanda. Ph.D. Dissertation, Department of Geography, Michigan State University.

Office National de la Population (ONAPO). 1990. Le Problème Démographique au Rwanda et le Cadre de la Solution Vol. I: Interrelations Population-Développement. Kigali: Rép. Rwandaise.

Office National de la Population (ONAPO). 1982. Actes du Colloques. Kigali: Rép. Rwandaise.

________. 1990. Le Problème Démographique au Rwanda et le Cadre de la Solution Vol. I: Interrelations Population-Développement. Kigali: Rép. Rwandaise.

République Rwandaise, 1992. Résumé de la Stratégie Nationale de l'Environnement au Rwanda et du Plan d'Action Environnemental. Kigali: Ministère de l'Agriculture, de l'Elevage et de l'Environnement, Service Nationale de l'Environnement.

Rwamiko. 1992. Transcription of recorded group interview of farmers. Rwamiko Commune, Gikongoro. August.

Schoenburg, David. 1993. We Are What We Eat: Ancient Agriculture Between the Great Lakes. Journal of African History 34: 1-31.

Schyns, Ch. 1988. Présentation de la Dissertation de J. Nzisabira "Evolution de l'Agriculture et Croissance de la Population au Rwanda". Bull. Séanc. Acad. r. Sci. Outre-Mer 33: 621-642.

Sirven, Pierre. 1984. La Sous-Urbanisation et les Villes du Rwanda et du Burundi. Ph.D. dissertation., Université de Bordeaux III.

Sorg, J.P. 1978. La Forêt Naturelle de Nyungwe (Rwanda): Exploitation ou Protection? Journal Forestier Suisse 129, 6: 445-452.

SNE (Service National de l'Environnement). 1992. Résumé de la Stratégie Nationale de l'Environnement au Rwanda et du Plan d'Action Environnemental (SNER/PAE). Kigali: Ministère de l'Agriculture, Service National de l'Environnement, Rép. Rwandaise.

Tardif-Douglin, Catherine, Serge Rwamasirabo and Katherine Krasovec. 1993. Statut nutritionnel et sécurité alimentaire au Rwanda: Résultats de l'enquête nationale sur la nutrition et la sécurité alimentaire des enfants de 0 à 5 ans et leurs mères. Division des Statistiques Agricoles Document du Travail No. 38. Kigali: MINAGRI.

Tiffen, Mary, Michael Mortimore and Francis Gichuki. 1994. More People, Less Erosion: Environmental Recovery in Kenya. New York, NY: John Wiley & Sons Ltd.

von Braun, Joachim, Hartwig de Haen and Juergen Blanken. 1991. Commercialization of Agriculture under Population Pressure: Effects on Production, Consumption, and Nutrition in Rwanda. Research Report 85. Washington, D.C.: International Food Policy Research Institute.

Uwizeyimana, Laurien. 1991. Croissance Démographique et Production Agricole au Rwanda: Impossible Adéquation? Cahiers du Cidep no. 8. Louvain-la-Neuve: CIDEP.

Michael J. Watts and Richard Peet (eds.). 1993. Economic Geography. Theme Issue: Environment and Development. Volume 69, Nos. 3, 4.

Weber, A.W. 1989. Conservation and Development on the Zaire-Nile Divide: An Analysis of Value Conflicts and Convergence in the Management of Afromontaine Forests in Rwanda. Ph.D. thesis, University of Wisconsin, Madison.

World Bank. 1993a. Rwanda: Note de Stratégie Economique vers une Croissance Durable. Document de Travail. Washington, D.C.: Département des Pays des Grands Lacs et de l'Océan Indien.

________. 1993b. Inventoring and Monitoring Biodiversity in the Nyungwe Forest. Mimeo. Washington, D.C.: The World Bank.


1 Current Address: Department of Geography, 1036 Derby Hall, 154 N. Oval Mall, Ohio State University, Columbus, OH 43210-1361 U.S.A. tel: 614-688-5758 fax: 614-292-6213


WHY BIODIVERSITY   FIELD STORIES   PARTICIPATE   RESULTS   MARKETSPACE

ABOUT THE NETWORK   WHAT'S NEW   LEARNING MATERIALS   SEARCH   LINKS   SITE MAP   HOME